[Bug] ITK ERROR: ITK only supports orthonormal direction cosines. No orthonormal definition found! (SimpleITK >= 2.1.0)

[alexandreroutier]

Hello everyone,

Thanks for this great tool :)

💀 Bug

When using nndet_prep [task] --full-check command, the following error occured:

File "<nnDetection>/nndet/utils/check.py", line 213, in _full_check
	img_itk_seq = [load_sitk(cp) for cp in case_paths]
File "<nnDetection>/nndet/utils/check.py", line 213, in <listcomp>
	img_itk_seq = [load_sitk(cp) for cp in case_paths]
File "<nnDetection>/nndet/io/itk.py", line 107, in load_sitk
	return sitk.ReadImage(str(path), **kwargs)
File "<CondaEnv>/lib/python3.9/site-packages/SimpleITK/extra.py", line 346, in ReadImage
	return reader.Execute()
File "<CondaEnv>/lib/python3.9/site-packages/SimpleITK/SimpleITK.py", line 8015, in Execute
	return _SimpleITK.ImageFileReader_Execute(self)
RuntimeError: Exception thrown in SimpleITK ImageFileReader_Execute: /tmp/SimpleITK-build/ITK/Modules/IO/NIFTI/src/itkNiftiImageIO.cxx:1980:
ITK ERROR: ITK only supports orthonormal direction cosines.  No orthonormal definition found!

This was noticed with the nnUNet framework when mixing different versions of SimpleITK an issue was raised (SimpleITK/SimpleITK#1433).

When using SimpleITK 2.0.2, this error does not occur. This seems to be due to recent changes with ITK when handling Nifti headers (see InsightSoftwareConsortium/ITK#2674 for further details and ongoing conversation).

I would think that freezing SimpleITK to SimpleITK < 2.1.0 would temporarily solve the issue.

Best

Environment

Environment Information:

----- PyTorch Information -----
PyTorch Version: 1.9.0
PyTorch Debug: False
PyTorch CUDA: 10.2
PyTorch Backend cudnn: 7605
PyTorch CUDA Arch List: ['sm_37', 'sm_50', 'sm_60', 'sm_61', 'sm_70', 'sm_75', 'compute_37']
PyTorch Current Device Capability: (7, 0)
PyTorch CUDA available: True

 


----- System Information -----
System NVCC: nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243

 

System Arch List: None
System OMP_NUM_THREADS: 1
System CUDA_HOME is None: True
System CPU Count: 6
Python Version: 3.9.6 (default, Jul 30 2021, 16:35:19)
[GCC 7.5.0]

 


----- nnDetection Information -----
det_num_threads 12
det_data is set True
det_models is set True```

[mibaumgartner]

Hi @alexandreroutier,

thanks for the detailed Issue, I'll fix the version of SimpleITK temporarily.
Did you encounter the error in the official examples or on a private data set?

Best,
Michael

[alexandreroutier]

Hi @alexandreroutier,

thanks for the detailed Issue with all the background info, I'll fix the version of SimpleITK temporarily.
Did you encounter the error in the official examples or on a private data set?

Best,
Michael

Indeed, I forgot to mention: this error was encounter on a private data set.

Best,
Alexandre

[mibaumgartner]

Hi @alexandreroutier ,

I updated the sitk version inside the requirements file for now.

I'm planning to rerun the preprocessing of the official data sets sometime in the future and will check if the latest version of sitk creates any Issues there. If it goes well I will probably unlock it / bump to the latest version again.

Best,
Michael

[pedrohbd]

Hi, same error here. It happens when I use the TorchIO library (that uses SimpleITK 2.1.1 internally). Cant change the SimpleITK version or the TorchIO wont work. Any solutions?

[mibaumgartner]

Hi @pedrohbd,
Can you use a different TorchIO version? If I remember correctly, the ITK version of TorchIO changes between releases. Otherwise you might need to look deeper into your data to fix it.

[malteekj]

I know this is old, but it is still a problem in simpleitk 2.2.0. After searching for a while, I realized the problem can stem from the qform_code and sform_code having different codes (qform is for using quaternion and sform is for using the affine matrix). A solution can be to play around with these and ensure that they conform with the definition of the coordinate system for your image.

A good resource to understand this better: https://brainder.org/2012/09/23/the-nifti-file-format/

[kabbas570]

@alexandreroutier the lower versions are not available, and still getting the same error.

[alexandreroutier]

Hi @kabbas570,

Which version of Python are you using? On my side, it was Python 3.7 / 3.8 and this workaround is working well so far. Maybe this can help with your specific installation.

In the meantime, it seems by the message of @malteekj on SimpleITK/SimpleITK#1433 that the error may still exist in latest version of SimpleITK.

Best,
Alex

[kabbas570]

@alexandreroutier thanks for your reply; I am using Python 3.9.7. Yes, it still exists, which limits the usage of SimpleITK for . nii files.

[machur]

This error does not occur anymore for SimpleITK version 2.2.1 (tested with Python 3.10). I upgraded from 2.1.1.2 as I described in issue MIC-DKFZ/HD-BET#25

[N1h1l1sT]

I'm actually experiencing the same issue with version 2.2.1 on Python 3.10:

Python version: 3.10.8 | packaged by conda-forge | (main, Nov 24 2022, 14:07:00) [MSC v.1916 64 bit (AMD64)]
SimpleITK version: 2.2.1

[machur]

I wish I could double-check the versions I specified, but, unfortunately, I don't have the data anymore to reproduce this issue.

[huangs-i2r]

data1.nii.gz
data2.nii.gz

Hi there, I have the same problem. My environment is python 3.8.11, SimpleITK 2.2.1.
Attached are two nii.gz files, data1 and data2, they are almost the same, but data1.nii.gz is OK to be opened by sitk.ReadImage() while data2.nii.gz will bump into the said error, a bit tricky.

[keesh0]

This seems to be fixed when I moved from Simple ITK 2.2.0 to 2.2.1 (under Python 3.8.3).
All is well.

[pangyuteng]

For those locked at py 3.6, you can try downgrading to v2.0 (pip install SimpleITK==2.0.0). I was able to load the nifti i had issue loading in v2.1.1.2.

[Wingseter]

In Simple ITK version 2.1.1.1 and 2.1.1, the following error occurred.
As mentioned by keesh0 , I changed the version to 2.2.1, and it worked without any issues.

[ThreeStones1029]

In Simple ITK version 2.1.1.1 , the same error occurred.
I changed the version to 2.2.0, and it worked without any issues.

[sydat2701]

I'm actually experiencing the same issue with version 2.2.1 on Python 3.10:

Python version: 3.10.8 | packaged by conda-forge | (main, Nov 24 2022, 14:07:00) [MSC v.1916 64 bit (AMD64)]
SimpleITK version: 2.2.1

Hi, I got the same problem. Did you solve it?

[keesh0]

It's been awhile since I last looked at this.
But upgrading from Simple ITK 2.2.0 to 2.2.1 (under Python 3.8.3) worked.

orthonormal direction cosines.

[DsLiuCase]

I replace my package，and using pip install SimpleITK==2.0.2. It works